Whenever you're ready, walk me through how you'd approach this.
Do you use CI/CD pipelines, and how do you deploy or integrate models to production?
Asked in the 1st Round Technical Interview stage. Explain the practical workflow from code and data validation through model release, serving integration, monitoring, and rollback. Discuss how your approach changes for batch inference versus online inference, and how you prevent training-serving skew and unsafe deployments.